Enhancing Early Diabetes Detection using Ensemble Machine Learning Models for Improved Prediction and Patient Care

Millions of individuals worldwide suffer from diabetes, a chronic condition that must be detected early to avoid major health issues like kidney failure, nerve damage, and heart disease. By using ensemble technique which combines machine learning (ML) approaches and evaluate the outputs and generates final prediction, then diabetes can be predicted more precisely. Random Forest, Support Vector Machine (SVM), and Logistic Regression, were the machine learning techniques and Voting Classifier is the ensemble classification model which were examined in this study. The study evaluated these model's ability to predict diabetes using datasets such as the Pima Indian Diabetes Dataset. Through the integration of Support Vector Machine (SVM), Random Forest, and Logistic Regression algorithms using a Voting Classifier, proposed model achieves an accuracy of 79%. This ensemble approach demonstrates superior performance compared to individual algorithms, with improved precision (0.81) and recall (0.78). It is identified that Predictive accuracy is found to be highly dependent on key parameters including glucose, BMI, and age with correlation coefficients of 0.47, 0.29, and 0.24 respectively. This method showed that combining multiple ML algorithms can significantly enhance diabetes diagnosis and care in addition to improving prediction capabilities. In the end, the study emphasizes how machine learning can completely change the way diabetes is treated and stresses how crucial it is to keep developing and using these technologies in order to improve patient outcomes.